Latent or low-level persistent reservoirs of HIV-1 may be the chief hurdle to eradication of infection. In particular, it is thought that latently infected T cells can harbor integrated virus, which cannot be eliminated by current therapeutics. Therefore if therapy is discontinued for any reason, this reservoir rekindles infection. Additional potential reservoirs, such as macrophages or microglia in the brain could serve as long-lived sources of low-level virus production. It is thus imperative to identify and model means to eliminate these reservoirs. This proposal aims to use an in Vivo model, the recently described "BLT mouse" to assess methods to attack latent reservoirs by an "induction/eradication" strategy. We will perform studies to test the concept that activation of latently infected cells, followed by introduction of specific agents designed to kill cells induced to newly express virus, will provide a means of purging these resevoirs in vivo. We will accomplish our goals through the following three Specific Aims: 1) Define the reservoirs of latent virus established by different HIV strains in BLT mice;2) Determine the effects of anti-HIV immunotoxin on latent reservoirs in vivo;3) Determine the capacity of genetically engineered cytotoxic T cells to impact viral reservoirs in vivo. The studies proposed herein will develop a valuable in vivo system to examine HIV latency, and may provide proof-of-principle that targeted immunoreagents may facilitate clearance of HIV reservoirs following induction of expression of the latent virus.